🎬 AI-Driven Movie Recommendation System

An end-to-end AI-powered movie recommendation system built using FastAPI, React, and Machine Learning.
The system provides personalized movie recommendations using Content-Based Filtering with TF-IDF and Cosine Similarity, deployed on modern cloud platforms.

---

🚀 Live Demo

• Frontend (Vercel):
  https://movie-recommendation-system-six-tawny.vercel.app/

• Backend API (Render):
  https://movie-recommendation-system-6gt8.onrender.com/docs

---

🧠 Key Features

• 🔐 JWT-based Authentication (Login & Register)
• 🎥 Browse top movies from TMDB dataset
• ⭐ Movie rating system
• 🤖 AI-powered movie recommendations
• 🧊 Cold-start handling (popular movies)
• ⏳ Lazy-loaded ML model (memory efficient)
• 🌐 Fully deployed (Frontend + Backend)
• 📱 Responsive UI (Desktop & Mobile)

---

🏗️ Tech Stack

Frontend

• React.js (Vite)
• Tailwind CSS
• Axios
• React Router
• Deployed on Vercel

Backend

• FastAPI
• REST APIs
• JWT Authentication
• MongoDB (users & history)
• Deployed on Render

Machine Learning

• Python
• Pandas, NumPy
• Scikit-learn
• TF-IDF Vectorization
• Cosine Similarity
• Lazy model loading for production

---

📊 Dataset

• Top 10K TMDB Movies Dataset
• Fields:
  • title
  • genre
  • overview
  • popularity
  • vote_average
  • vote_count

---

🤖 How Recommendation Works

1. Movie overviews are converted into TF-IDF vectors
2. Cosine Similarity finds similar movies
3. Top-N similar movies are returned
4. Model loads only when /recommend endpoint is called (lazy loading)

---

📡 API Endpoints

Method	Endpoint	Description
POST	/auth/register	Register user
POST	/auth/login	Login user
GET	/movies	Get movies
GET	/recommend/{movie}	Get AI recommendations
POST	/rate-movie	Rate movie
GET	/user/history	Watch history

---

🔐 Environment Variables

Backend uses environment variables for security.

MONGO_URI=your_mongodb_uri SECRET_KEY=your_secret_key

---

🚀 Deployment

• Backend deployed on Render
• Frontend deployed on Vercel
• CORS configured for cross-domain access
• Optimized for free-tier hosting (single worker, lazy ML load)

---

🧪 Challenges Solved

• Cold-start latency on ML endpoints
• CORS issues between Vercel & Render
• Dataset path resolution in production
• Memory optimization for ML models

---

📌 Future Enhancements

• Collaborative Filtering
• Hybrid Recommendation System
• Movie posters using TMDB API
• User-specific recommendations
• Admin dashboard

---

👨‍💻 Author

Amit Chaurasia
B.Tech | Full-Stack & AI Enthusiast